The International Monitoring System (IMS) is comprised of multiple types of sensors that provide verification information. While each piece of information is useful for verification, the full benefit of multi-technology measurements has not been fully taken advantage of. Data Fusion is an approach that seeks to integrate disparate sources of data into a unified and comprehensive event analysis. Several approaches (e.g. cost-function analysis, Bayesian inference) have demonstrated the power and benefit of data fusion approaches for Treaty verification. However, an important problem in the data fusion process arises when not all information is consistent, or believable. Dempster-Schafer theory provides a statistical means to reconcile evidentiary beliefs in the data fusion process. This poster will describe how inconsistent evidence may arise within the IMS, and show how Dempster-Schafer theory can help to reconcile evidence in a data fusion process and support the event analysis process for National Data Centres.
This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344. This abstract is LLNL-ABS-817217-DRAFT.
Uncertainty characterization is crucial in data fusion processes (e.g. inference techniques) that combine evidence from multiple sources. When pieces of evidence are inconsistent, applying Dempster-Schafer theory reconciles the inconsistencies and improves the inference process.